Query Result API

[namedgraph]

I think it the query Result API is also awkward af.

There is a bit of magic here that makes this appear like different Python objects, depending on the type of result.
If the type is “SELECT”, iterating will yield lists of ResultRow objects
If the type is “ASK”, iterating will yield a single bool (or bool(result) will return the same bool)
If the type is “CONSTRUCT” or “DESCRIBE” iterating will yield the triples.

I personally hate this sort of "magic". I would prefer to have less of it, not more, and make things explicit. For example, that the method signature would be

query(query_object ...) -> Union[Graph, Result, bool]

where Graph is for DESCRIBE/CONSTRUCT results, Result is for SELECT results only, and bool is for ASK results.

Right now having a Result object there is no good way (that I know of) to determine what form of query it is coming from.

[ajnelson-nist]

This topic is being visited in Issue 2283, and development in progress on PR 2320, though stalled due to availability.

If there are questions about API style preference, my own preference is to defer to definitions or model suggestions from the upstream sources. I haven't reviewed SPARQL's specification recently enough to recall if there is a Result object suggested. But please, if we can find an upstream source instead of attempting to resolve opinions at only the style layer, let us defer and cite.

[namedgraph]

There are 2 forms of SPARQL results: RDF graph/triples (Graph in RDFLib) and SPARQL Query Results (with XML, JSON, CSV formats). The latter does not currently have a distinct representation in RDFLib IMO, as all forms of results are conflated into Result.

ASK result is a special case since it's a single boolean, but it's specified as a case of SPARQL Query Results.
Some engines (e.g. Jena) support RDF dataset/quad results (Dataset in RDFLib) as an extension beyond SPARQL 1.1